Adrian Grange [Fri, 3 May 2013 16:59:54 +0000 (09:59 -0700)]
Merge "Extend number of reference buffers to 8." into experimental
Adrian Grange [Thu, 2 May 2013 23:41:19 +0000 (16:41 -0700)]
Extend number of reference buffers to 8.
The number of reference buffers is extended to 8 and
a reference sign-bias added for the LAST_FRAME.
Whilst the number of reference buffers used by an
individual frame remains unchanged at 3, these may
now be selected from 8 possible buffers.
Change-Id: I2d247b9c1c2b3a339d6c9fac125e81ba373f75a7
Scott LaVarnway [Fri, 3 May 2013 14:34:29 +0000 (07:34 -0700)]
Merge "Removed is_inter_mode() check" into experimental
Scott LaVarnway [Fri, 3 May 2013 14:30:31 +0000 (07:30 -0700)]
Merge "Reduced y_dequant, uv_dequant size" into experimental
Ronald S. Bultje [Fri, 3 May 2013 04:03:38 +0000 (21:03 -0700)]
Fix use of wrong rate/distortion variables in 16x8 r/d check.
Change-Id: Ib5961b4c8ca84d54c84b2651a4e0317c72fe7da4
Dmitry Kovalev [Fri, 3 May 2013 03:48:36 +0000 (20:48 -0700)]
Merge "Removing MAXF macro and using MAX instead." into experimental
James Zern [Fri, 3 May 2013 02:32:03 +0000 (19:32 -0700)]
Merge changes Ifea8618a,I014b832a into experimental
* changes:
convolve_test: cosmetics
convolve_test: remove unnecessary memset
James Zern [Thu, 2 May 2013 20:08:19 +0000 (13:08 -0700)]
convolve_test: cosmetics
fix indent, whitespace, casts
Change-Id: Ifea8618a90f9da263a8955dd242bb3aa7fc59ae5
John Koleszar [Fri, 3 May 2013 01:53:27 +0000 (18:53 -0700)]
Merge "superframe_test: use delete[] where appropriate" into experimental
James Zern [Fri, 3 May 2013 01:14:38 +0000 (18:14 -0700)]
superframe_test: use delete[] where appropriate
Change-Id: Id374267c93a7e14e985b8079833364c8eff5248b
Ronald S. Bultje [Thu, 2 May 2013 23:40:57 +0000 (16:40 -0700)]
Fix 16x16-iteration indexing bug in main encode_sb_row loop.
With this, encoder/decoder appear to match with sb8x8 experiment.
Needs some larger-scale testing.
Change-Id: I44d3cac37b3c98264985ed0a0fc763c30089aa64
Jingning Han [Thu, 2 May 2013 21:38:01 +0000 (14:38 -0700)]
Fix state update in sb8x8 rate-distortion loop
Update mode_info of 8x8 blocks within the scope of current block.
Change-Id: I110c599e60664a5acde6afd919b107cea8419a0d
Jingning Han [Thu, 2 May 2013 21:32:57 +0000 (14:32 -0700)]
Merge "Cache splitmv motion vectors appropriately in decoder." into experimental
Ronald S. Bultje [Thu, 2 May 2013 20:56:06 +0000 (13:56 -0700)]
Merge "Fix i4x4 mode reading and writing in sb8x8 bitstream." into experimental
Ronald S. Bultje [Thu, 2 May 2013 20:55:08 +0000 (13:55 -0700)]
Cache splitmv motion vectors appropriately in decoder.
This fixes some problems in splitmv-related bitstream parsing in the
sb8x8 experiment.
Change-Id: Id168f9ec9c88626877d8f536e6f394b6f4f782e4
Ronald S. Bultje [Thu, 2 May 2013 20:00:33 +0000 (13:00 -0700)]
Fix i4x4 mode reading and writing in sb8x8 bitstream.
Don't allow i4x4 except for sb8x8 recursion step. Read only 4 (not 16)
i4x4 submodes if we are i4x4.
Change-Id: Iaaaced1a134006b2c96eed66f014300eae41e0ed
Ronald S. Bultje [Thu, 2 May 2013 19:35:59 +0000 (12:35 -0700)]
Merge "Set transform size correctly for non-tx_select splitmv/i4x4_pred." into experimental
James Zern [Thu, 2 May 2013 19:29:34 +0000 (12:29 -0700)]
convolve_test: remove unnecessary memset
input_ is filled with random values just afterward.
the size was wrong anyway as input_ is allocated with memalign so
sizeof(input_)==sizeof(uint8_t*)
Change-Id: I014b832ac60960cd22b6f369dbc9fd648d4055b5
Ronald S. Bultje [Thu, 2 May 2013 19:26:00 +0000 (12:26 -0700)]
Set transform size correctly for non-tx_select splitmv/i4x4_pred.
Fixes mismatch in keyframes with sb8x8 experiment enabled.
Change-Id: Id44c25e1cd32be5c6543ed8cd840a5b0c8e726f1
Dmitry Kovalev [Thu, 2 May 2013 18:57:16 +0000 (11:57 -0700)]
Removing MAXF macro and using MAX instead.
Change-Id: I51c53692b1150005645bf362c5e5a8275178a8fd
Ronald S. Bultje [Thu, 2 May 2013 18:10:07 +0000 (11:10 -0700)]
Merge "Fix intra4x4 decoder reconstruction stride for sb8x8 experiment." into experimental
Ronald S. Bultje [Thu, 2 May 2013 18:00:35 +0000 (11:00 -0700)]
Fix intra4x4 decoder reconstruction stride for sb8x8 experiment.
Change-Id: I5a886a6fa19086629a274f9ada7a862fd0ddf62f
Ronald S. Bultje [Thu, 2 May 2013 17:53:51 +0000 (10:53 -0700)]
Merge "Always initialize values in sb/mb_partitioning." into experimental
Scott LaVarnway [Thu, 2 May 2013 17:44:33 +0000 (13:44 -0400)]
Removed is_inter_mode() check
If a reference frame is inter, the only valid modes would
be inter modes. This check is unnecessary.
Change-Id: Ib8433ab5a3418f94149ee4e3062d48d7740d225a
Dmitry Kovalev [Thu, 2 May 2013 17:40:58 +0000 (10:40 -0700)]
Merge "Using treed_read/treed_write functions for segment ids." into experimental
Ronald S. Bultje [Thu, 2 May 2013 17:34:37 +0000 (10:34 -0700)]
Always initialize values in sb/mb_partitioning.
This doesn't affect the output, because in previous cases where the
values were uninitialized, this was because the mb_row/col is outside
the codable area, and thus encode_sb will test them for the next
decomposition-level, but return right after that on size-check. All
this does is prevent a warning in valgrind.
Change-Id: I90d8a29e6f8ebb2b0143684e08fe77ae3a0816b1
Ronald S. Bultje [Thu, 2 May 2013 17:15:24 +0000 (10:15 -0700)]
Fix right-edge availability for intra prediction in sb8x8.
Fixes valgrind uninitialized value use warnings.
Change-Id: Ie9314d684e2ad194f8aca5bde1729fb9b7c0221d
Ronald S. Bultje [Thu, 2 May 2013 16:08:33 +0000 (09:08 -0700)]
Merge "Fix some more offset errors in sb8x8." into experimental
Jingning Han [Thu, 2 May 2013 16:04:28 +0000 (09:04 -0700)]
Merge "Fix bug in sb8x8 partition context" into experimental
Ronald S. Bultje [Thu, 2 May 2013 14:54:18 +0000 (07:54 -0700)]
Fix some more offset errors in sb8x8.
Change-Id: I83677227f7610fdf2db9f15f87fecd4d8e072427
Ronald S. Bultje [Thu, 2 May 2013 14:11:36 +0000 (07:11 -0700)]
Merge "Fix block reconstruction with sb8x8 enabled." into experimental
Jingning Han [Thu, 2 May 2013 05:26:08 +0000 (22:26 -0700)]
Fix bug in sb8x8 partition context
Fix the issue that causes array bound excess in getting partition
context.
Change-Id: I66166f047f0bcaefebb0bcf441c5b1f777d8da44
Jingning Han [Thu, 2 May 2013 03:52:07 +0000 (20:52 -0700)]
Merge "Fix bugs in sb8x8 experiment/context prob update" into experimental
Ronald S. Bultje [Thu, 2 May 2013 02:28:17 +0000 (19:28 -0700)]
Fix block reconstruction with sb8x8 enabled.
The encoder reconstruction is now correct. Decoder to follow shortly.
Change-Id: Iedf98cdaebb4ca1256c7714cad7024a75853ad6a
Ronald S. Bultje [Thu, 2 May 2013 02:26:26 +0000 (19:26 -0700)]
Merge "Fix bug in sb8x8 bitstream decoding." into experimental
Ronald S. Bultje [Thu, 2 May 2013 00:26:46 +0000 (17:26 -0700)]
Merge "Store splitmv modes in context after 8x8 rd loop." into experimental
Ronald S. Bultje [Wed, 1 May 2013 21:50:39 +0000 (14:50 -0700)]
Fix bug in sb8x8 bitstream decoding.
Change-Id: I7ac69a369851b09d9422f2e3c62b12b3fcd49a4c
Ronald S. Bultje [Wed, 1 May 2013 21:43:33 +0000 (14:43 -0700)]
Store splitmv modes in context after 8x8 rd loop.
Change-Id: I07aa89a67e0ac5f99ef0c448553dbc46b0ed27f2
Jingning Han [Wed, 1 May 2013 21:35:07 +0000 (14:35 -0700)]
Fix bugs in sb8x8 experiment/context prob update
Fix bugs occur in contextual partition probability update, when
sb8x8 is enabled.
Change-Id: I19e2cec8a54c2dafd2be2803bbfde7337a2ae45f
Johann [Wed, 1 May 2013 22:13:42 +0000 (15:13 -0700)]
Merge "Remove unused quantize optimizations." into experimental
James Zern [Wed, 1 May 2013 21:58:36 +0000 (14:58 -0700)]
Merge "Change to fourcc to enable decoding of ffmpeg files." into experimental
Ronald S. Bultje [Wed, 1 May 2013 21:45:27 +0000 (14:45 -0700)]
Fix some crashes in sb8x8 experiment.
Change-Id: I390bb1cedc835f439fd5dd6cda6572b29cbb139c
Jingning Han [Wed, 1 May 2013 20:48:14 +0000 (13:48 -0700)]
Merge "Enable bit-stream support to SB8X8" into experimental
Scott LaVarnway [Wed, 1 May 2013 20:25:10 +0000 (16:25 -0400)]
Reduced y_dequant, uv_dequant size
Currently, only two values are used. Removed the unused
values.
Change-Id: Idc5b8be354d84ffc68df39ea3e45f9f50d977b35
Jingning Han [Wed, 1 May 2013 16:43:59 +0000 (09:43 -0700)]
Enable bit-stream support to SB8X8
This commit enables bit-stream writing and reading for recursive
partition down to block 8x8.
Change-Id: I163cd48d191cc94ead49cbb7fc91374f6bf204e2
Dmitry Kovalev [Wed, 1 May 2013 19:04:21 +0000 (12:04 -0700)]
Merge "Adding vp9_get_qindex function." into experimental
John Koleszar [Wed, 1 May 2013 04:43:26 +0000 (21:43 -0700)]
Merge "Make vp9_optimize_sb* common" into experimental
John Koleszar [Tue, 30 Apr 2013 16:54:51 +0000 (09:54 -0700)]
Make vp9_optimize_sb* common
Unify the various vp9_optimize_sb functions into one that handles all
transform sizes.
Change-Id: I48b642fbfb3e72cc2e0bcf1d0317a80a80547882
Jingning Han [Tue, 30 Apr 2013 19:08:32 +0000 (12:08 -0700)]
Migrating MB decoding into decode_sb/atom
The decode_mb only carries I8X8_PRED decoding, which will be covered
by the regular MB intra modes when SB8X8 is on. To be removed later.
Change-Id: I3b9ee55917a30b42518b81987bc10c22b1a19e7f
Johann [Wed, 1 May 2013 01:42:05 +0000 (18:42 -0700)]
Remove unused quantize optimizations.
Files were copied from vp8 and never maintained.
Change-Id: I9659a8755985da73e8c19c3c984423b6666d8871
Dmitry Kovalev [Wed, 1 May 2013 00:49:55 +0000 (17:49 -0700)]
Merge "Cleaning up encoder segmentation code." into experimental
Ronald S. Bultje [Wed, 1 May 2013 00:41:57 +0000 (17:41 -0700)]
Merge "Use more restrictive block radius for 8x8 block MV references." into experimental
Ronald S. Bultje [Wed, 1 May 2013 00:34:02 +0000 (17:34 -0700)]
Use more restrictive block radius for 8x8 block MV references.
Change-Id: If02e006aa8a89da9de23da92362bd2e7718ea07c
Ronald S. Bultje [Wed, 1 May 2013 00:29:46 +0000 (17:29 -0700)]
Merge "sb8x8 integration in rd loop." into experimental
Ronald S. Bultje [Tue, 30 Apr 2013 23:13:20 +0000 (16:13 -0700)]
sb8x8 integration in rd loop.
Work-in-progress, not yet ready for review. TODO items:
- bitstream writing (encoder) and reading (decoder)
- decoder reconstruction
Change-Id: I5afb7284e7e0480847b47cd0097cb469433c9081
John Koleszar [Tue, 30 Apr 2013 22:34:58 +0000 (15:34 -0700)]
Merge "Expand UMV border to 96 pixels" into experimental
Dmitry Kovalev [Tue, 30 Apr 2013 21:06:49 +0000 (14:06 -0700)]
Using treed_read/treed_write functions for segment ids.
Changing the order of probabilities inside mb_segment_tree_probs in order
to use treed_read/treed_write function instead of custom code.
Change-Id: I843487d5057913b9358db73da270893eefecc6c8
Dmitry Kovalev [Tue, 30 Apr 2013 20:39:50 +0000 (13:39 -0700)]
Adding vp9_get_qindex function.
Moving common code from encoder and decoder to vp9_get_qindex function.
Also moving quant-related constants from vp9_onyxc_int.h to
vp9_quant_common.h.
Change-Id: I70c5bfbaa1c8bf00fde0bfc459d077f88b6d46c8
Yaowu Xu [Tue, 30 Apr 2013 19:26:09 +0000 (12:26 -0700)]
Merge "Removed code no longer being used." into experimental
Jingning Han [Tue, 30 Apr 2013 19:17:56 +0000 (12:17 -0700)]
Merge "Migrating SPLITMV_4X4 decoding into decode_atom" into experimental
Yaowu Xu [Tue, 30 Apr 2013 19:09:27 +0000 (12:09 -0700)]
Removed code no longer being used.
Change-Id: Iab9a88f250614a790b6ad96bf3150a74210910df
Dmitry Kovalev [Tue, 30 Apr 2013 19:00:35 +0000 (12:00 -0700)]
Merge "Adding vp9_update_frame_size function." into experimental
Jim Bankoski [Mon, 29 Apr 2013 21:24:21 +0000 (14:24 -0700)]
Change to fourcc to enable decoding of ffmpeg files.
Change-Id: Ia16f167d904d53bc38cb3a8c841c6c52daad500f
(cherry picked from commit
296421d84cd1ffd074ff3fe4a772238c897f1bda)
reverts: dbd050c vpxdec: correct VP[89] fourccs
Dmitry Kovalev [Tue, 30 Apr 2013 18:14:27 +0000 (11:14 -0700)]
Adding vp9_update_frame_size function.
Moving common code from encoder and decoder to vp9_update_frame_size.
Change-Id: I6ca758b7d05ffd52821bd3f7ad68089da11e4165
Dmitry Kovalev [Tue, 30 Apr 2013 18:07:05 +0000 (11:07 -0700)]
Merge "General cleanup inside vp9_bitstream.c." into experimental
Jingning Han [Tue, 30 Apr 2013 17:16:09 +0000 (10:16 -0700)]
Migrating SPLITMV_4X4 decoding into decode_atom
Separate the decoding process of 4x4 block based coding (both intra
and inter) from decode_mb and move it into decode_atom_. This allows
to further move the rest per 16x16 block decoding of decode_mb into
decode_sb, and hence eventually deprecating decode_mb when SB8X8 is
enabled.
Change-Id: I678cb8007d8a57b792d7a23020edb0c74fbf4237
Dmitry Kovalev [Tue, 30 Apr 2013 17:21:24 +0000 (10:21 -0700)]
Merge "Renaming refresh_entropy_probs to refresh_frame_context." into experimental
Dmitry Kovalev [Tue, 30 Apr 2013 17:19:48 +0000 (10:19 -0700)]
Merge "Consistent names for quant-related functions and variables." into experimental
John Koleszar [Tue, 30 Apr 2013 15:43:07 +0000 (08:43 -0700)]
Merge "Fix incorrect 32x64, 16x32 subpel variance" into experimental
Jingning Han [Tue, 30 Apr 2013 04:51:59 +0000 (21:51 -0700)]
Merge "Separate I4X4_PRED coding from macroblock modules" into experimental
Jingning Han [Mon, 29 Apr 2013 19:43:38 +0000 (12:43 -0700)]
Separate I4X4_PRED coding from macroblock modules
Separate the functionality of I4X4_PRED from decode_mb. Use
decode_atom_intra instead, to enable recursive partition of superblock
down to 8x8.
Change-Id: Ifc89a3be82225398954169d0a839abdbbfd8ca3b
Ronald S. Bultje [Tue, 30 Apr 2013 00:38:25 +0000 (17:38 -0700)]
Make sure 32x32 coefficients fit in int16_t.
Fixes mismatches on yt.
Change-Id: I65444137578b29a4984af5c01b2874c5ab086acc
John Koleszar [Mon, 29 Apr 2013 20:44:54 +0000 (13:44 -0700)]
Expand UMV border to 96 pixels
Ensures that the full 64 pixel border is available for prediction (need a minimum of
64+INTERP_EXTEND on all sides, and 32+INTERP_EXTEND on UV). Value also must be a
multiple of 32 to keep UV stride alignment. The smaller border was causing the prediction
to read outside the frame, which can cause a mismatch.
TODO: Get rid of this explicit border and use edge emulation instead.
Change-Id: I3f68453a088ec0ab4349d0f5cc02b573be06d7c4
Yaowu Xu [Mon, 29 Apr 2013 23:16:27 +0000 (16:16 -0700)]
Merge "fixed new intra code for rectanglar blocks" into experimental
Dmitry Kovalev [Mon, 29 Apr 2013 23:07:17 +0000 (16:07 -0700)]
Cleaning up encoder segmentation code.
Moving code from vp9_pack_bitstream to new function encode_segmentation.
Change-Id: I1f1e59a1f038618ad95162b7db4b6f8164850ea8
Yaowu Xu [Mon, 29 Apr 2013 23:05:19 +0000 (16:05 -0700)]
Merge "Enabled i4x4 to use right above pixels" into experimental
Yaowu Xu [Mon, 29 Apr 2013 23:02:35 +0000 (16:02 -0700)]
Merge "Use same intra prediction for all block size" into experimental
Yaowu Xu [Mon, 29 Apr 2013 22:49:00 +0000 (15:49 -0700)]
fixed new intra code for rectanglar blocks
Also fixed two minor subtle boundary conditions in intra prediction
code, and replaced memcpy/memset with vpx_ prefixed version.
Change-Id: I9cddff3be831228b628f1f2f065a61feacbcbee6
Yaowu Xu [Fri, 26 Apr 2013 00:52:09 +0000 (17:52 -0700)]
Enabled i4x4 to use right above pixels
Change-Id: I7442b4600b6812bed13e655ccf68f9ea56cc83a2
Yaowu Xu [Tue, 23 Apr 2013 18:06:11 +0000 (11:06 -0700)]
Use same intra prediction for all block size
The commmit changed to use same intra prediction function for all
block sizes.
Some details on the changes:
1. All directional modes except DC/TM/V/H now have built-in filtering
for all pixels with filter taps either (1, 2, 1)/4 or (1, 1)/2.
2. Above edge get automatic extended to double width (bw*2), which
makes a lot of the prediciton mode computation simpler.
3. Same intra prediction function is called with different size
for i4x4_pred and all other larger size.
Overall, the change helped keyframe only coding for both cif size
and std-hd size test sets by .5% consistently on all encodings.
For normal coding with single/auto key frame, the change now also
is consistently net positive for all encodings. The overall gains
is about .15% on std-hd set.
Change-Id: I01ceb31fbc73d49776262e6bdc06853b03bbd1d1
John Koleszar [Mon, 29 Apr 2013 19:59:30 +0000 (12:59 -0700)]
Fix incorrect 32x64, 16x32 subpel variance
Incorrect strides caused use of uninitialized stack data.
Change-Id: Ifc12174285311bcf0453987d4283e91141f979ef
John Koleszar [Mon, 29 Apr 2013 19:53:04 +0000 (12:53 -0700)]
Merge "Use foreach_transformed_block with tokenize_b" into experimental
Ronald S. Bultje [Mon, 29 Apr 2013 19:41:55 +0000 (12:41 -0700)]
Merge "Merge encode_sb and encode_sb64 in a recursive encode_sb/b function." into experimental
Ronald S. Bultje [Mon, 29 Apr 2013 19:41:53 +0000 (12:41 -0700)]
Merge "Fix mi_row argument to update_stats()." into experimental
Ronald S. Bultje [Mon, 29 Apr 2013 19:41:50 +0000 (12:41 -0700)]
Merge "Factor out assigning correct partitioning seg_context pointers." into experimental
Ronald S. Bultje [Mon, 29 Apr 2013 19:41:47 +0000 (12:41 -0700)]
Merge "Use recursive function in resetting txfm_size for skip-blocks." into experimental
Ronald S. Bultje [Mon, 29 Apr 2013 19:33:18 +0000 (12:33 -0700)]
Merge "Use recursive function in choosing segmentation map coding method." into experimental
John Koleszar [Mon, 29 Apr 2013 18:57:48 +0000 (11:57 -0700)]
Use foreach_transformed_block with tokenize_b
Updates the tokenizer to use the common block walker used by the
detokenizer, to support non-4:2:0 and more than 3 planes.
Change-Id: If1854117a9c7c1427349209fa2b3051ce6459dcb
Ronald S. Bultje [Sun, 28 Apr 2013 15:59:13 +0000 (08:59 -0700)]
Merge encode_sb and encode_sb64 in a recursive encode_sb/b function.
Change-Id: Ifd1f7319e398381ca7d55aaacc68c39a1319d651
Ronald S. Bultje [Sat, 27 Apr 2013 22:59:37 +0000 (15:59 -0700)]
Fix mi_row argument to update_stats().
This doesn't change output, because the argument isn't actually used
ATM. However, we should fix it for consistency.
Change-Id: I7b7326a8e92c0d411c999ec2c781204b516ed53d
Ronald S. Bultje [Sat, 27 Apr 2013 22:52:22 +0000 (15:52 -0700)]
Factor out assigning correct partitioning seg_context pointers.
Change-Id: I6e787915a92e0abdea86adaa3e1106aa9c165034
Ronald S. Bultje [Sat, 27 Apr 2013 22:26:39 +0000 (15:26 -0700)]
Use recursive function in resetting txfm_size for skip-blocks.
This is shorter than fully writing out all cases (52 vs. 84 LOC), and
is easier to extend.
Change-Id: Ic75f86f15cf037543c3dc6234d9ac088205a0ed1
Ronald S. Bultje [Sat, 27 Apr 2013 22:14:42 +0000 (15:14 -0700)]
Use recursive function in choosing segmentation map coding method.
This is shorter than fully writing out all cases (55 vs. 84 LOC), and
is easier to extend.
Change-Id: Ibad1bf6463198cad9188c5507181d4b03d393e6d
Jingning Han [Mon, 29 Apr 2013 18:28:56 +0000 (11:28 -0700)]
Merge "Immigrate tokenize_mb into tokenize_sb" into experimental
Ronald S. Bultje [Mon, 29 Apr 2013 18:28:25 +0000 (11:28 -0700)]
Merge "Change above/left_context to use an 8x8 basis." into experimental
Jingning Han [Mon, 29 Apr 2013 18:05:22 +0000 (11:05 -0700)]
Immigrate tokenize_mb into tokenize_sb
Unify the tokenize_ function and enable configurable block size for
superblock 8x8. We are immigrating the functionalities of
macroblock handles into superblock ones, and eventually will remove
encode_mb and decode_mb. To be continued on detokenize_ module.
Change-Id: I9f81e8c2291082535cf5e0c4b662eb24fb7c8a7f
Ronald S. Bultje [Mon, 29 Apr 2013 17:37:25 +0000 (10:37 -0700)]
Change above/left_context to use an 8x8 basis.
Output changes slightly because of a minor bug in (at least) the sb32x16
block2above tx16x16 tables that previously existed in vp9_blockd.c.
Change-Id: I624af28ac200a8322d64454cf05c79e9502968cc
Deb Mukherjee [Mon, 29 Apr 2013 17:13:06 +0000 (10:13 -0700)]
Merge "Turning model based reverse update on for coefs" into experimental
Deb Mukherjee [Fri, 26 Apr 2013 18:13:26 +0000 (11:13 -0700)]
Turning model based reverse update on for coefs
Turns model based reverse updates on for coefficients in an
effort to reduce the memory requirement for counters.
With this patch the counters needed will be reduced by about
75% since only 3 counts are needed instead of 12.
The impact in performance is:
derf300: -0.252%
stdhd250: -0.046%
However retraining should alleviate some of the drop in
performance.
Change-Id: I6f2b3e13f6d5520aa3400b0b228fb5e8b4a43caa
John Koleszar [Sat, 27 Apr 2013 00:52:35 +0000 (17:52 -0700)]
Unify decode_sb for >16x16 in vp9_decodframe.c
First patch to make sb decoding based on the transform size. This patch
is working for the sb modes, combining the parts of decode_mb that fit
into this framework will come as a second patch.
Change-Id: I26123416a7a87e096bbdb5eb944ce5bb198384f8